LED Drive cct design...advice wanted

Hi,

This cct originally started from a desire to use a PIC to do both brightness control (via PWM) and the comparators to drive a mosfet/inductor current/voltage source...all to reduce component count...but it got shelved for other things.
I've seen a recent nixie tube PSU implementation that has recently brought this idea back, and the recent Vf drop over time thread.

Instead of using a PIC, I thought a simple analog cct may work. Basically I want to create control signals to drive a current limiter cct by measuring not only the current, but the slope of Vf/If curve (I'll explain further down).

The reason being is to somehow keep the LED at its optimal operating point...regardless of the effects of heat and heatsinking. To simplify, a 1W LED without a heatsink could be set to be driven at 350mA....but probably shouldn't. A well heatsinked LED could.

I'm intending to put small Vf spikes across the LED (deltaVf), then filter the current sense resistor to get a delaIf and use that to control the drive current....not a constant current set point.

In theory, the If-Vf curve will shift with temperature, and after reading the Vf-time shift thread with time as well. A high deltaIf indicates the LED is operating on the steep part of the curve....a low deltaIf indicates the LED could be driven higher. You could set the driver to work at a specified deltaIf...rather than a specified If. Thus this cct could keep the LED in the optimal and most reliable operating point chosen, regardless of heatsinking/age issues.

Questions are:
-the deltaIf/deltaVf ratio can also be used to run the LED at any defined point in the Vf/If curve......whats a good point though? I'm presuming the midpoint of the bend.
-Any similiar designs out there you've heard of?
-Any nasty effects of putting small spikes in the LED would presumably b a function of their duration and size, how short or small would they have to be before damage would occur?
-This cct would increase reliability....brightness would be expected to shift with time.....any guestimates on the amount if an LED is kept at a specific part of its Vf/If curve.

I can't answer your questions with any reliability, but the concept seems hypothetically sound, not to mention fascinating !

I can't wait to hear some of our local experts weigh in on this one :thinking:

Thanks,

I actually had a thought just after reading this. Instead of a constant deltaVf being generated somehow, it would be an easier implementation to use an adder cct on the current feedback loop to induce a constant If spikes....then an opamp could read the Vf to determine the deltaVf via a filter.

You could set the operating point by a set point at a specified deltaVf which would be a correspond to a point on the Vf/If curve (ie constant Vf/If ratio, if deltaIf kept constant in the adder).

Phil
(I've a feeling in the hairs on the back of my neck, someone has suggested this before....???)